Week 6 (Correlation)

Question 1

What is a correlational design

Answer 1

-We look at pairs of scores to see whether one measure is consistently associated with scores on another measure.

Question 2

Ordinal data

Answer 2

-Categories that can be ordered/ranked - property of magnitude
-BUT no precise differences between ranks
-So, although the categories have an order, they might not evenly spaced
E.g, above average, average, below average poor

Question 3

Linear relationships

Answer 3

A relationship between two variables that can be described by a straight line

Question 4

Outlier

Answer 4

Extreme data point, clear odd one out

Question 5

What other things can scatterplots tell us

Answer 5

Spot restricted ranges
-When range is restricted , correlation can go down
Subgroups
-Scatterplots can show us whether we have a false positive correlation due to different subgroups of participants.

Question 6

Pearson’s Product Moment Correlation: r

Answer 6

ranges from -1 (perfect negative correlation) to +1 (perfect positive correlation)
-Parametric test so it’s out “powerful” correlation - check this one first
-Must meet certain assumptions, including normal distribution of data

Question 7

Directional problem

Answer 7

The direction of the causal relationship may not be what we assume

Question 8

Confound problem

Answer 8

There may be another variable that isn’t measured in the study, but causes both shirt size and academic achievement

Question 9

Spearman’s Rank Correlation: rho p

Answer 9

-Non-parametric test
-Can be used with ordinal data , or continuous data that are not normally distributed

Question 10

Kendall’s Tau Rank Correlation: tau t

Answer 10

-Often used instead of spearman’s when there are tied scores (same scores across multiple people)

Question 11

Assumptions of Pearson’s r

Answer 11

Type
-Data should be continuous, rather than ordinal (though often we treat Likert data as continuous - we’ll come back to judging when this is appropriate)

Normal
-Both variables should approximate a normal distribution (no strong skew, etc.)

Extreme?
-There should be no extreme values (outliers)
-Outliers can overly influence the calculation f Pearson’s statistic, more so than the other data points - leads to an inaccurate result.

Question 12

Assumptions of Spearman’s ranked correlation

Answer 12

Type
-Data must be ordinal , interval or ratio level of measurement

Ties
-Participants variable levels should not be the same across multiple people (there should be few tied scores)

Question 13

Covariance

Answer 13

The extent to which variables co-vary (change together)
-High covariance: Means there is a large overlap between the patterns of change (variance) observed in each variance
-Low overlap: means there is little overlap in the variance of each variable